On Sun, 21 Jan 2001, Flip ter Biecht wrote:

> As we're all getting younger by the day, let me confirm that the problem
> with winmodems is that there's no comport in them. External modems use an
> existing comport and should be able to react to a valid signal sent through
> that comport. (It does however not mean that every external modem is hayes
> compatible...)

  To expand... it's actually the UART chip which winmodems are 
missing. (Universal Asynchronous Reciever Transmitter)  This chip 
sends signals *through* the serial port.  It isn't a "comport" 
itself.

  An internal modem has its own UART onboard.  An external modem
uses the UART on the motherboard which is associated with the 
serial port in use, and of course a "winmodem" / "software modem" / 
"modem emulator" simply has no UART at all.

  Instead of using a hardware UART, it uses a software UART 
emulator, which puts load on your CPU that should rightfully be 
delegated a separate hardware UART.  (Bill Gates refers to this
as "using excess CPU cycles"... as if IE is such a lightweight
app that it leaves the CPU sitting their idling... ehem)

  Winmodems generally only ship the "drivers" for Windows, though
I have seen a winmodem or two with drivers for DOS included
(a few years back).  I told my father that if the box says "UART 
16550" on it, it's a real modem.  He replied that the modem he 
was looking at didn't say that, but it did say it would run DOS
programs.  I then mistakenly told him that if it would run DOS 
programs, it must certainly be a "real" modem.  He bought it.  
I was wrong.  It was one of those rare winmodems with a DOS 
driver.  :-(

  It would be *possible* for manufacturers to create a software 
UART emulator (driver) for any OS.  Of course, the whole reason for 
winmodems is to reduce cost, and putting more effort into drivers 
for little-used OS's would erode the winmodem profit margins... or
raise the price to the point that it would be cheaper to buy 
a modem with a hardware UART.

> What, by the way are standard io addresses and irq values for internal pci
> modems? (I noticed a new help page in arachne 1.69 at
> file://doc\pcimodem.htm, which just gives an example with io d800 irq 10.

  I *believe* that most PCI devices are plug'n'play.  They may
have default values built-in, but I think they're set up by the
BIOS if the motherboard is PNP aware.  Otherwise, set-up by
the PNP OS.  Failing either of those, then you'll need some kind
of setup utility to write the desired values... anyway this is as 
I recall it from trying to make a PNP sound card work on Linux and 
DOS from some years back.  The state of Linux pnp was still quite
rudimentary at the time.

  In order to use sound, I was booting to DOS (pnp setup was 
called from autoexec.bat IIRC), which wrote the desired values to
the card.  Then I booted Linux using loadlin, and the values 
written while in DOS remained intact. 

> It took me months lately to find out the address of a pci ethernet card at
> 0x6100 irq 11 -or in fact redhat setup found it for me- and I'd be  most
> happy with a more complete list of pci devices because the luxury of
> pentium, pci and pnp is still rather new in the recycling stores where I
> usually compose my systems)

  In Linux, you can do '/sbin/lspci -v' to find out all kinds
of info on your PCI bus and the cards on it... but this will only
tell you one possible combination of several.  I guess in many 
instances that would be enough?

  Here's the kind of stuff lspci will tell you:

00:09.0 Ethernet controller: Lite-On Communications Inc LNE100TX (rev 20)
        Subsystem: Netgear FA310TX
        Flags: bus master, medium devsel, latency 64, IRQ 11
        I/O ports at e800
        Memory at ea000000 (32-bit, non-prefetchable)
        Expansion ROM at e9000000 [disabled]

  So, if you're setting this up in DOS, you know at least e800
and IRQ11 are valid values.  
 
> Apart from that, I found another nasty bit of hardware: a Compaq Qvision
> pci video card (1 mb, from 1996), that is autodetected nicely by win95, but
> not by Arachne. 

  Compaq has a reputation for "innovative design" or in the 
words of the end user trying to figure it out, "deviating from
the standard."  

  The couple of Google links I looked at don't seem too promising:
http://www.xfree86.org/pipermail/newbie/2000-November/002479.html
http://support.microsoft.com/support/kb/articles/Q111/5/70.ASP

-- 
Steve Ackman                                      [EMAIL PROTECTED]
Glass Host, Arts & Crafts                  http://www.delphi.com/crafts
Metamorphosis Glassworks Page      http://twovoyagers.com/metamorphosis



Reply via email to